Risk-adjustable stochastic schedule based on Sobol augmented Latin hypercube sampling considering correlation of wind power uncertainties

The risks associated with wind power forecast (WF) deviations are of paramount importance to many power system participants (PSPs). However, traditional sampling approaches are computationally prohibitive to model these deviations. Additionally, setting a risk level for satisfying different PSPs receives little attention. This paper constructs a risk-adjustable stochastic day-ahead scheduling (RSDS) model to balance the risk requirements of PSPs, and proposes a Sobol-augmented Latin Hypercube Sampling (SaLHS) approach to improve sampling efficiency for scenario generation process in RSDS. At first, SaLHS and D-vine copula are combined to generate WF error scenarios for RSDS considering correlations of wind farms. Specifically, SaLHS improves the uniformity and removes the correlation of random samples. Then, a Glue-VaR-based generation adequacy index (GVGAI) is proposed to measure operational risk. By adjusting the parameters of GVGAI, a desirable risk level can be obtained considering requirements of different PSPs. Furthermore, a multi-objective RSDS model is constructed considering operational cost and GVGAI. At last, an entropy-Weighted Aggregated Sum Product Assessment method is proposed to find the best compromise solution for RSDS model based on the Pareto front obtained by an epsilon-constraint method. A modified IEEE-RTS system is used to validate the effectiveness of proposed method via numerical simulations.

Automated summary

This paper presents a risk-adjustable stochastic day-ahead scheduling model for balancing the risk requirements of PSPs and proposes an improved sampling approach. By combining SaLHS and D-vine copula, WF error scenarios can be generated to account for wind farm correlations. A Glue-VaR-based generation adequacy index is proposed to measure operational risk and adjust risk levels based on PSP requirements.